Sujeet Gund
Agentic AI Engineer · M.Tech AI @ VIT Bhopal
Building autonomous agentic systems, production RAG pipelines, and LLM-powered infrastructure.
About
Engineering intelligent systems from first principles to production
Final-year Integrated M.Tech in AI at VIT Bhopal, currently building production AI systems at Divam Technologies.
I specialize in agentic architectures — LangGraph multi-agent workflows with HITL checkpoints, hybrid RAG with pgvector, and FastAPI backends on Google Cloud Run. The kind of systems that handle real workloads, not just demos.
Outside work, I ship ambitious side projects: an email processing SaaS, an anonymous geo-social platform, and tooling that pushes LLM orchestration further than most tutorials go.
Work Experience
GenAI Developer Intern
Led GenAI development across real-time voice AI, multi-agent graph workflows, and enterprise RAG systems from concept to production AWS deployment.
- Architected multi-agent state graphs using LangGraph to automate complex decision workflows in Publie.ai — implemented intent routing (<200ms decision latency) and HITL checkpoints yielding an 85%+ autonomous resolution rate.
- Engineered real-time voice & chat agents integrating Meta Cloud API, LiveKit, and Plivo for OmniAgent — built resilient webhook pipelines with BullMQ queues guaranteeing sub-second response latency and zero-drop event delivery under high concurrent (~1200rps) traffic.
- Built context-aware lead capture agent for Newton On Mars using hybrid RAG over live site content — accelerated client technical discovery and automated PRD draft generation (~70% reduction in onboarding time).
Featured Projects
GroundedAI — Self-Correcting Multi-Source Agentic RAG
A self-evaluating agentic RAG platform powered by LangGraph, Reciprocal Rank Fusion (RRF) pgvector + tsvector hybrid search, and local zero-cost DeBERTa v3 NLI faithfulness scoring to eliminate hallucinations (94.2% grounding precision).
rzp Merchant — AI-Native E-Commerce & Agentic Commerce (MCP/ACP)
An autonomous e-commerce platform implementing Model Context Protocol (MCP) and Agentic Commerce Protocol (ACP), allowing external AI agents to discover, basket-optimize, and execute transactions via Razorpay under 3-tier security mandates.
MailMind — Agentic Email Processing & HITL Orchestration Engine
An event-driven email processing agent built with LangGraph and Resend Inbound APIs. Features intent classification (96.5% accuracy), RAG auto-drafting, and PostgreSQL-persisted Human-in-the-Loop (HITL) approval checkpoints yielding an 88% operational time reduction.
Skills
Languages & Databases
AI & ML
Backend & Infrastructure
Certifications
Complete Data Science, Machine Learning, DL, NLP Bootcamp
Issued by Udemy
Applied Machine Learning in Python
Issued by University of Michigan
Supervised Machine Learning: Regression and Classification
Issued by DeepLearning.AI
AWS Cloud Essentials
Issued by Amazon Web Services (AWS)
Education
VIT Bhopal University
Sep 2023 - Mar 2028
Integrated M.Tech in Artificial Intelligence
CGPA: 9.33